Abstract: To solve the problems of particle degradation and the limited accuracy of battery state of charge (SOC) estimation of the particle filter algorithm (PF), a joint estimation algorithm combining the online parameter identification of extended Kalman (EKF) at macro time scale and the improved particle filter algorithm (LOCR-UKPF) state estimation at micro time scale based on second-order RC equivalent circuit was studied. UKPF and PF models, and the simulation verification of the algorithms was carried out under the conditions of the Federal City Timetable (FUDS) and Highway Timetable (US06). The simulation results show that the Root Mean Square Error (RMSE) of the EKF-LOCR-UKPF algorithm considering the time scale, importance density function and resampling strategy is reduced by 21.6%, 30.7% and 47.0% compared with the LOCR-UKPF, UKPF and PF algorithms, respectively, and the Root Mean Square Error (RMSE) is reduced by 36.9%, 43.8% and 55.4% under the US06 condition, respectively. The improved EKF-LOCR-UKPF joint estimation algorithm has improved the estimation accuracy of battery SOC, and has certain application value and prospects in power battery SOC prediction and battery management.